Anthropic is introducing a research preview of 'Claude Code auto mode,' set to launch no earlier than March 12, 2026. This new permission setting allows the AI coding assistant to autonomously handle permission requests during development sessions. The goal is to enable developers to run longer, uninterrupted tasks without constant manual approvals.
This move addresses a common pain point for developers who often resort to less secure methods, like the `--dangerously-skip-permissions` flag, to avoid workflow disruptions. While auto mode offers enhanced convenience, Anthropic cautions that it's not foolproof and recommends its use in isolated environments. Users should also anticipate a slight increase in token usage, cost, and latency.
By default, auto mode will be accessible to users by running the command 'claude --enable-auto-mode.' No specific team-wide enablement action is required from administrators.
For organizations wishing to restrict access, disabling auto mode can be achieved through two primary methods. MDM or OS-level policies can be configured: on macOS, by setting the 'disableAutoMode' key to 'disable' within the com.anthropic.claudecode managed preferences domain; on Windows, by setting the same key in the HKLM\SOFTWARE\Policies\ClaudeCode registry path.
Alternatively, file-based managed settings offer another control layer. Administrators can create or modify a settings file, setting the 'disableAutoMode' key to 'disable.' The typical file locations are /Library/Application Support/ClaudeCode/managed-settings.json on macOS, /etc/claude-code/managed-settings.json on Linux/WSL, and C:\Program Files\ClaudeCode\managed-settings.json on Windows.
